Understanding Spatial Skills and Encoding Strategies in Student Problem Solving Activities

Abstract

Background and Context. Margulieux's Spatial Encoding Strategy theory (SpES) provides a possible reason for the relationship between spatial skills and success in STEM fields, including CS. While there is indirect evidence to suggest that the theory holds, there is little work which explicitly explores the core theory in practice. Furthermore, current work in spatial skills has largely focused on introductory courses, and it is unclear whether advanced students (and then experts) use spatial skills in computing. Objectives. We wish to determine whether we can see senior students in CS with high spatial skills utilising non-verbal encoding strategies when solving CS programming problems. Method. Transcripts from a think-aloud exercise with experienced students (final year of undergraduate), whose spatial skills were measured, were analysed to identify utterances which indicated spatial encoding strategies being employed, such as the construction and alteration of mental models on the fly, and to determine differences according to spatial skills level. Findings. Students with higher spatial skills were more likely to exhibit evidence of the construction of flexible, comprehensive mental models to solve the programming problems, demonstrating advanced encoding and chunking strategies. Students with lower spatial skills were more likely to struggle with the construction and alteration of mental models, indicating that they typically lack the capability to effectively chunk and save working memory space. Implications. This work confirms the predictions of SpES more precisely than prior work by showing that skilled problem solving involves the mental model creation and manipulation that underlies SpES. It demonstrates that students with better spatial skills are more likely to succeed in programming problem solving, even in the later stages of study, due to their ability to encode non-verbal information.